Directory UMM :Data Elmu:jurnal:I:International Journal of Educational Management:Vol13.Issue5.1999:
Effects of stay-back on teachers' professional
commitment
Antia Hung
St Teresa Secondary School, Kowloon, Hong Kong
James Liu
Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
Keywords
Teachers, Commitment, Schools,
Management
Abstract
This study investigated the effects
of stay-back on teachers' professional commitment, in order to
develop an effective means to
measure teachers' commitment to
the teaching profession which,
according to various studies, contributes to ``quality education''.
Other possible factors, such as
age, gender, marital status, tenure, educational level and position, were also studied.
Suggestions on ways to increase
commitment, based on the statistical findings, were also made. The
results of causal-comparative,
correlation and multiple regression analyses revealed that stayback was the one factor which
was found to be the most highly
and significantly related to commitment. Apart from educational
stay-back, marital status, age and
tenure were also proved to relate
significantly to commitment. In
this research, the accuracy and
reliability of the stay-back data
were improved through adjustment for a secondary analysis.
The International Journal of
Educational Management
13/5 [1999] 226±240
# MCB University Press
[ISSN 0951-354X]
[ 226 ]
In Hong Kong, the study of teachers' professional commitment is of the utmost importance. People begin to question the
effectiveness of the school, especially when
juvenile delinquency seems to become a
serious problem and when academic performance seems to have fallen in the wake of
rapid expansion of secondary and tertiary
education. The situation is further aggravated by the declining teacher quality as a
result of teacher wastage.
Teacher wastage is an obstacle to quality
education. Since the 1980s, the teaching
profession in Hong Kong has been facing the
problem of teacher exodus as a result of the
political uncertainty in Hong Kong after 1997.
Government statistics show that wastage of
school teachers has been edging up over the
past few years. In secondary schools, the
wastage rate among teachers has increased
from 6 per cent in 1988 to more than 10 per
cent annually (South China Morning Post,
1990). Research carried out by the Hong Kong
Institute of Personnel Management (HKIPM)
in 1994 revealed that the education sector
provided the second highest number of
professional people leaving the territory in
the preceding two years and the number of
teachers emigrating showed a big increase.
School managers have difficulty in maintaining some committed teachers who have
been in the profession for quite some time
and this in turn makes it difficult for schools
to become effective.
The problem that the education field faces
is further exacerbated by the fact that fewer
people are willing to join the teaching
profession. From 1987-90, applications for
full-time teacher training have dropped by
about 40 per cent. This was attributed to the
lack of promotion prospects in schools,
difficult students and demanding parents, the
burden of too many non-professional duties
The current issue and full text archive of this journal is available at
http://www.emerald-library.com
and the expansion of tertiary education
which the majority of the students in Hong
Kong will put as the first priority of their
educational pursuit. After all, teaching is
regarded as a stressful job which is becoming
more unpopular and, therefore, it is less
preferred by the young in Hong Kong.
On the other hand, there is the student
problem. With the advances of society, juvenile problems have been escalating. There
are many contributing factors, for example,
material temptation, broken families, pressure from school work, the inclusion of all
school-age children in compulsory education,
etc. The new generation of students are more
rebellious and quite a number of them are
trouble-makers. They do not only do poorly
in their academic work but they also have
behavioural problems. Facing such students,
the teaching profession demands more commitment from teachers. But unfortunately,
the supply of committed teachers falls far
short of demand. Wishing to teach the cream
of the crop is the general mentality of the
teachers. Very few are willing to spare extra
time and effort to help the under- or even
average-achievers.
With reference to the situation in Hong
Kong, to arouse the commitment of teachers
is a pressing problem befalling all school
managers nowadays. It has been reported
that teachers' commitment was positively
related to leader's charisma (Bass, 1985;
Cheng, 1990). The Hong Kong Education
Department is trying to introduce reforms in
school management. One typical example is
the school management initiative (SMI) under which schools are given more flexibility
and autonomy in resource management
(Education and Manpower Branch and Education Department, 1991). Another example is
the recent introduction of quality education
fund to finance proposals which aim at
improving the quality of education (Education and Manpower Branch and Education
Department, 1998). The essence of these
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
schemes is participation and accountability.
It is believed that if teachers are involved in
the setting of school goals and the decisionmaking process, they will tend to be committed members of staff. Arousing the commitment of teachers is thus one of the
objectives of these schemes.
This paper presents an effective and economical means which can truly predict
commitment and yet is free from the limitations of the conventional methods of measuring commitment that meet with
employees' suspicion of the motives of the
management, as this new approach can be
applied without employees' knowing it. It
focuses on investigating the correlation of
teachers' stay-back to their professional
commitment.
Commitment and contributing
factors
By professional commitment, we mean teachers' commitment to their job as a teacher.
It corresponds to commitment to work put
forward by Fink (1992) which is related to
organization, co-worker and work.
Since teaching is a profession comprising a
considerable number of members, conceptually we can see the teaching profession as a
large organization. Based on an analogy to
organization commitment defined by Mowday and McDade (1979), we characterize
professional commitment by the following
factors:
.
a strong belief in and acceptance of the
profession's goals and values;
.
a willingness to exert considerable effort
on behalf of the profession; and
.
a strong desire to maintain membership
in the profession.
Numerous studies have suggested that employees' job attitudes and commitment can
affect their performance (Cammann et al.,
1983; Oldham and Hackman, 1981). For personal factors, they include locus of control
(Cheng, 1990), age, tenure, gender and education level (Fink, 1992). Non-demographic
individual difference variables include central life interest (Dubin et al., 1975), desire for
greater job responsibility, expectations about
goal achievement and personal needs (Steers,
1977).
For organizational characteristics, they
consist of job, leadership, work group and
organizational characteristics. Among these
four characteristics, leadership and organizational characteristics seem to be of utmost
concern to school managers who want to
arouse the commitment and enhance the
performance of teachers. Leadership refers to
the styles of the leaders which can be supervisory (Fukami and Larson, 1984), charismatic (Conger et al., 1988), one of initiating
structure and consideration (Morris and
Sherman, 1981) and one of giving rewards
(Bateman and Strasser, 1984). For organizational characteristics, they include things
like formalization, decentralization, functional dependence on the work of others and
participation in decision making (Morris and
Steers, 1980; Steers and Rhodes, 1978).
Other factors such as lateness, absenteeism, tardiness and turnover are the manifestations of commitment as reflected in most
literature. But there is a kind of behaviour
which can be a predictor of employees' level
of commitment and which has not yet been
studied thoroughly. It is ``employees' stayback time after normal office hours''. As
discussed before, leadership style is an
important factor affecting commitment. By
providing teachers with inspiration, encouragement and more meaning to their work, a
charismatic principal can enhance a teacher's faith in and respect for him and these
lead to an increase in the teacher's commitment to the principal and so to the school and
the work. Additional studies by Cole (1979)
and Sekaran (1989) have further reinforced
our belief that employees' ``stay-back'' is also
a manifestation of commitment and can be
used as a predictor of commitment level,
apart from the traditional commitment-related behaviour mentioned by other researchers.
In short, it is suggested that a school
manager can generate commitment from his
employees by building trust, letting people
develop their own ways of working, sharing
accountability and ownership of a job, and
negotiating help and supervision in terms
that engender employee development. But
above all, managers need to be role models
for their subordinates by being committed
and they should also empower others in their
jobs and roles.
Traditional measures of
commitment
There are two approaches to measure commitment: formal and informal. The formal
approach refers to the administering of
questionnaires to employees to measure their
commitment. Two commonly-used questionnaires are the ``organizational commitment
questionnaire'' devised by Mowday and
McDade (1979) and the ``commitment diagnosis instrument'' by Fink (1992). However, the
[ 227 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
formal method is often met with resistance
from the employees, especially in an organization where there is a lack of trust from the
management. Employees may suspect the
motives of the management. Instead of being
viewed as a means of understanding the
organization and the needs of its employees,
the questionnaire can easily be considered as
a form of employee evaluation that leads to
judgments about the quality of the employees
themselves, which can form the basis for
promotion, demotion or dismissal. As a
result, employees exercise extra caution, not
because they understand that they must be
honest but because they are more concerned
about how they should respond to the questions. Even attempts to guarantee anonymity
can be futile. The results generated from the
formal method may not be a true picture of
employee commitment.
The informal approach employs the observational techniques like commitment indicators which are actually commitmentrelated behaviour, for instance, absenteeism,
lateness, turnover, etc. Although absenteeism is the most widely studied predictor, it is
not an accurate and reliable one as most
reviewers of the literature on absenteeism
argue that current research abounds in
problems of inadequate theory, inaccurate
measurement, invalid data and faulty analytic procedures (Goodman et al., 1984).
Stay-back
In Hong Kong, all the aided secondary
schools are full-time. Recesses and lunch
time are very short, totalling less than two
hours each day. Teachers have a very heavy
workload. On average, a teacher has to teach
six periods out of eight every day. Apart from
teaching, teachers also have to perform some
non-teaching duties, for example, they may
have to be on duty during recess or lunch
time. So, they have little contact with students and colleagues during school hours.
Counselling work, extra-curricular activities, the marking of assignments, socializing, etc. will have to be carried out after
school hours. Without staying back, it is
quite unlikely that teachers can accomplish
their tasks and perform well.
As a commitment predictor, stay-back can
be defined as the time that employees stay
behind in the office after the formal and
scheduled office hours, regardless of what
they do (Cole, 1979; Sekaran, 1989). In this
study, the employees referred to the aided
secondary school teachers in Hong Kong
randomly selected for this research purpose
[ 228 ]
and stay-back was broadly divided into two
types, namely voluntary and compulsory
stay-back. Voluntary stay-back happens out
of the teachers' own free will. On the other
hand, compulsory stay-back, as the name
suggests itself, happens because teachers are
required to stay as part of the fulfilment of
their official duties.
In addition, voluntary stay-back can
further be divided into two sub-groups:
educational stay-back and non-educational
stay-back. Figures 1 and 2 give the classifications of the various stay-back activities.
Data collection
The present study was conducted among
aided secondary school teachers in Hong
Kong. In the first stage, questionnaires were
distributed directly to about 5 per cent of
aided secondary schools in Hong Kong randomly chosen for this research. A return rate
of 45 per cent was recorded. However, among
these questionnaires, one third did not include complete data, thus making analysis
difficult and were then discarded. In order to
improve the reliability of the study, another
batch of questionnaires was given out in the
second stage at the Professional Teachers'
Union where teachers used to go shopping.
Teachers were randomly intercepted and
teachers in government aided and subsidized
secondary schools were asked to fill in the
questionnaire. The response rate was quite
reasonable and both stages managed to
collect a total useful sample of 1,220 representing about 5 per cent of teacher population
in aided secondary schools.
In the questionnaire, teachers were asked
questions about their general feelings towards the teaching profession, matters pertaining to their stay-back, and personal
information such as age, gender, marital
status, position and teaching experience.
Figure 1
Classifications of activities in compulsory stayback
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
Figure 2
Classifications of activities in voluntary stay-back
The International Journal of
Educational Management
13/5 [1999] 226±240
Teachers' frequency of stay-back and the
total stay-back hours per month were calculated from the information that they gave in
the questionnaire. The dependent variable
(teachers' professional commitment) was regressed on the independent variables (stayback indices, age, gender, marital status,
tenure and position). In addition, the characteristics of stay-back data were analysed
and investigated.
Sample design and analysis
The sample consists of 1,220 teachers in Hong
Kong who were randomly chosen from some
aided secondary schools and from those who
visited the Professional Teachers' Union. In
terms of gender, the subjects were quite
equally distributed, 48 per cent male and 52
per cent female. The average age of those in
the sample was 35, with the 21-30 age group
accounting for 42 per cent, the 31-40 group 46
per cent and the 41-50 group 12 per cent. Fiftysix per cent of the subjects were single and 44
per cent were married.
As far as education is concerned, 65 per
cent of the subjects held bachelor's degrees or
above while 35 per cent were non-graduates
who might only have a teacher's certificate.
In Hong Kong, the non-graduates start
teaching as Certificated Masters/Mistresses
(CM), then they may be promoted to assistant
masters/mistresses (AM). Some outstanding
performers may then be further promoted to
senior assistant masters/mistresses (SAM)
and finally to principal assistant masters/
mistresses (PAM), which, at present, is a
position held by one only in every aided
secondary school with an average of 50
teachers (see Figure 3).
In the stream for graduates, they start as
graduate masters/mistresses (GM). Some
may be promoted to senior graduate masters/
mistresses (SGM). Further promotion will
make them end up as principal graduate
masters/mistresses(PGM) who are actually
deputy principals. At present, there are two
PGMs in every aided secondary school.
Among the sample in the present study, 28
per cent are CMs, 5 per cent AMs, 42 per cent
GMs, 22 per cent SGMs, and 3 per cent PGMs,
i.e. 31 per cent of the respondents had been
promoted to senior positions. When teaching
experience is concerned, 35 per cent had
taught for five years or less while those who
had taught between six to ten years constituted 31 per cent, 11 to 15 years 25 per cent, 1620 years 7 per cent and only 2 per cent had
taught for more than 20 years (see Figure 4).
The questionnaire documented the frequency and the duration of and the reasons
for the teachers' stay-back. All respondents
were asked to put down on average how often
they stayed back each month and the average
Figure 3
Distribution of professionals
[ 229 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
Figure 4
Distribution of teaching experience
The International Journal of
Educational Management
13/5 [1999] 226±240
amount of time for each stay-back. Some
typical stay-back activities, which can
broadly be divided into two types: voluntary
and compulsory, were also listed. These data
can be used to cross check the total amount of
time a respondent spent on the stay-back
activities in an average month, regardless of
the nature of stay-back, be it voluntary or
compulsory. To prevent any patterns that
might lead a subject to automatically circle
numbers down one side or the other of the
seven-point scale used in the questionnaire,
some of the items in the question set are
stated positively, where a higher score reflects greater commitment; while some of the
items are stated negatively, where a higher
score reflects less commitment. Also the
positive and negative statements were mixed
randomly, not systematically, again to avoid
any pattern. Items were summed and converted to a 10-100 scale to generate the
commitment index for each subject.
Stay-back indices
The study employed six stay-back indices as
follows:
1 Total stay-back index by frequency estimate (TSIFE).
2 Total compulsory stay-back index (TCSI).
3 Total educational stay-back index (TESI).
4 Total non-educational stay-back index
(TNSI).
5 Total voluntary stay-back index (TVSI).
6 Total stay-back index (TSI).
The TSIFE was generated by multiplying the
total number of stay-back days each month
with the average amount of time for each
stay-back, regardless of the reasons for stayback, where the number of stay-back days
each month was calculated from the frequency (once how many days) of stay-back,
assuming 20 school days per month.
The TCSI only counted those activities
which had been classified as compulsory
stay-back activities in the last section. Since
the respondents had specified the number of
times they carried out a certain compulsory
[ 230 ]
stay-back activity in an average month and
the duration each time, the total amount of
time spent on a certain activity in a month
can be found. The TCSI is actually a summation of all the time spent on compulsory
stay-back activities by a respondent in a
month.
The TESI counted those activities classified as voluntary educational stay-back activities. Again the total hours spent on these
activities by a respondent in an average
month was a summation of the time spent on
each voluntary educational stay-back each
month which was calculated by multiplying
the number of times with the duration each
time. As discussed in the last section, some of
the activities listed in the questionnaire can
actually be grouped under either educational
or non-educational. In the data analysis, half
of the total amount of time spent on this
group of activities was counted in the TESI
while the other half of it was counted in the
TNSI.
The TNSI counted those activities defined
as voluntary non-educational stay-back activities. It is the summation of the total
amount of time spent on each of these
activities by a respondent in an average
month which was calculated by multiplying
the number of times with the duration each
time. Like the TESI, the TNSI also included
half of the total amount of time spent on
activities which can be counted as either
educational or non-educational. The TVSI is
the sum of the TESI and the TNSI. The TSI is
the sum of the TCSI, TESI and TNSI.
In theory, the TSIFE should equal the TSI if
the respondents were absolutely consistent
in answering the questionnaire. The built-in
cross-checking device during data collection
is a means to ensure reliability of data.
Hypotheses testing
Basically, the present research examined
some hypotheses using statistical techniques
such as differences in means and standard
deviations, t-test, F-test correlations, and
multiple regression model.
Before testing the hypotheses, the consistency and reliability of the stay-back data
generated from frequency estimate and that
obtained by an aggregation of the total
compulsory stay-back hours, total educational stay-back hours and total non-educational stay-back hours were investigated. The
two kinds of information were related to each
other and their correlation coefficient was
calculated.
Apart from the above method, the following was also employed in the data analysis.
Stay-back data were analysed by each stay-
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
back index to examine whether or not its
sample distributional characteristics approximate normality. The same has been
done for variables like age, position and
tenure. Specifically, Kendall and Stuart's
criteria were used to examine normality of
sample distributions. Those independent
variables with normal distributional patterns could then be utilized in the multiple
regression analysis.
Hypothesis 1
Teachers who stayed back voluntarily were
distinguished from those who did not by
forming two groups. Using the total voluntary
stay-back hours (both educational and noneducational), the first group was formed by
subjects with total voluntary stay-back hours
less than the median, while the second group
was subjects with total voluntary stay-back
hours more than the median. Then, the means
and the standard deviations of the commitment indices of the two groups were calculated.
A t-test was conducted to see the significance of
the difference between the two groups and the
correlation coefficients between commitment
and stay-back were compared.
Hypothesis 2
Teachers who stayed back voluntarily were
divided into two groups according to the
nature of their stay-back. The first group
consisted of those who spent 51 per cent or
more of the total voluntary stay-back time on
educational stay-back activities while the
second group comprised those who spent 51
per cent or more of their total voluntary stayback time on non-educational stay-back activities. Then, the means and the standard
deviations of the commitment indices of the
two groups were calculated. A t-test was
conducted to see the significance of the
difference between the two groups and the
correlation coefficients between commitment
and educational stay-back and that between
non-educational stay-back were compared.
Hypotheses 3, 4 and 5
Teachers were divided into two groups
according to their educational level, gender
and marital status, as required in Hypotheses
3, 4 and 5 respectively. In Hypothesis 3, the two
groups of teachers were graduates and nongraduates. In Hypothesis 5, teachers who were
single and those who were married made up the
two groups. Under each hypothesis, the means
and the standard deviations of the commitment indices of the two groups were calculated.
A t-test was conducted to see the significance of
the difference between the two groups. The
correlation coefficients between commitment
and educational level, as in Hypothesis 3, that
between commitment and gender, as in
Hypothesis 4 and that between commitment
and marital status, as in Hypothesis 5 were
compared.
Hypotheses 6-14
Teachers' professional commitment was predicted by working out the correlation coefficient between each subject's commitment
and each of the independent variables such
as TVSI, TESI, TNSI, age, gender, marital
status, tenure, position and educational level.
This was presented in the form of a correlation matrix. A line of best fit, y = bx + a, was
generated, to facilitate the necessary prediction, when and where the relationship was
significant and amply correlated.
Hypothesis 15
The few most influential predictor variables
were used for a 3-4 step multiple regression
analysis. The resulting multiple correlation
coefficients (R) were examined for their
magnitudes.
Statistical analysis
Findings of the analysis of the hypotheses
include two main streams: the examination
of the stay-back data and the prediction of
professional commitment using various independent variables. The primary concern of
the first stream was to test stability of the
stay-back data along with other statistical
attributes (i.e. normality) associated with the
six stay-back indices. The second stream
focused on validating the use of voluntary
stay-back indices and other independent
variables like age, gender, tenure and educational level to predict professional commitment.
Consistency and reliability of stay-back
indices
It was found that TSIFE and TSI were
correlated with each other and gave a
correlation of 0.48, significant at p = 0.01
level. Therefore, it can be concluded that the
two indices are highly and positively correlated and the respondents were consistent in
giving their stay-back data. Since the other
stay-back indices such as TCSI, TESI and
TNSI are complements of the TSI, like the TSI
and TSIFE, they are reliable indices for
statistical analysis in the present research.
Normality of the stay-back indices
Hypotheses 1 to 5 were tested by means of
causal-comparative method. The t-test is
probably the most commonly used statistical
tool in causal-comparative studies and it has
made three assumptions about the scores
[ 231 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
obtained in research. One is that the scores
in the population under study are normally
distributed. So, it was necessary to test the
normality of the stay-back data before any
statistical analysis was carried out.
The characteristics of each sample distribution were examined using measures of
skew and kurtosis, as well as pictorial
shapes. Statistically, distributions may be
considered non-normal, indicating the presence of skew and leptokurtosis, when the
skew approaches 2 and kurtosis exceeds 5. In
normal distribution, both skew and kurtosis
are zero. If the value of the skew is positive,
distribution is skewed to the left but if its
value is negative, the distribution is skewed
to the right. For kurtosis, its value determines the peak of the distribution. If kurtosis
is greater than zero, the distribution is more
peaked than normal but if kurtosis is less
than zero, the shape of the curve is flatter
than the normal curve.
In this study, all six stay-back indices were
found to deviate from normal distribution.
According to Kendall and Stuart's criterion
for normality, the distributions of TCSI and
TNSI seemed to be furthest from normal. As
shown in Table I, kurtosis and skew of the
former are 3.88 and 1.77 respectively while
that of the latter are 3.2 and 1.83 respectively.
The other four stay-back indices, namely
TESI, (kurtosis = ±0.04 and skew = 0.83), TSI,
(kurtosis = ±0.59 and skew = 0.4), TVSI,
(kurtosis = ±0.27 and skew = 0.74) and TSIFE,
(kurtosis = ±1.23 and skew = ±0.2) more
closely approximated normality. In all the
four indices, the distribution curve is flatter
than the normal one as the values of kurtosis
are less than zero and the sample distribution tends to be skewed to the left in the cases
of TESI, TSI and TVSI which means that the
majority of the sample had stay-back hours
less than the mean.
Although the six stay-back indices do not
have perfectly normal distribution, they are
within an acceptable range of normal distribution with different magnitudes of deviations. They, therefore, are fit for a t-test
analysis.
[ 232 ]
In reporting the results of the analysis, the
significance level was taken at p 4 0.1. It was
proposed that teachers who stay back are
more committed than those who do not. The
analysis revealed that the mean commitment
index of teachers who stayed back was a bit
higher than that of teachers who did not (72.2
vs 72). Nevertheless, the difference was not
statistically significant (see Table II).
The analysis for the hypothesis that teachers who stay back more for educational
purposes are more committed than those who
stay back more for non-educational purposes
was found to be infeasible because there was
a great difference in the number of subjects
between the two groups of teachers. Out of
the 1,220 subjects, only 61 (i.e. 5 per cent)
stayed back more for non-educational purposes, while 1,139 (i.e. 93.4 per cent) have
longer stay-back hours for educational purposes than non-educational ones. Twenty
respondents (i.e. 1.6 per cent) reported no
stay-back of any kind. So, a comparison could
not be made.
For the non-graduates who were posited to
be more committed teachers than graduates,
results showed that the mean commitment
index of graduate teachers was 73.34 and that
for non-graduate teachers was 71.64. This
finding contradicted the hypothesis but such
a finding is not statistically significant (see
Table III).
It was proposed that women are more
committed teachers than men. When the
commitment indices of the two groups were
analysed, the mean commitment index of
men was a bit higher than that of women
(72.20 Vs 72.04). But the t-test demonstrated
that the two groups were not different from
each other in terms of commitment at a
statistically significantly level (see Table IV).
Table II
Commitment of teachers who stay back and
teachers who do not
Variable
Table I
Sample distribution of various independent variables
Mean
Minimum
Maximum
Standard deviation
Kurtosis
Skewness
N
Results
Commitment
TCSI
TNSI
TESI
TSI
TVSI
TSIFE
10.84
0.00
48.50
9.32
3.88
1.77
1220
2.89
0.00
15.43
3.6
3.2
1.83
1220
27.22
0.00
74.00
19.23
±0.04
0.83
1220
40.95
1.90
87.00
21.29
±0.59
0.40
1220
30.11
0.00
74.00
20.18
±0.27
0.74
1220
35.06
0.91
60.00
17.36
±1.23
±0.20
1220
Teachers Teachers not
stay-back
stay-back
(Mean)
(Mean)
72.20
72.00
t
p
0.57 >0.1
Table III
Commitment of graduate teachers and nongraduate teachers
Variable
Commitment
Graduate Non-graduate
teachers
teachers
(Mean)
(Mean)
73.34
72.00
t
p
±0.3 > 0.1
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
Table IV
Commitment of male teachers and female
teachers
Variable
Commitment
Male
teachers
(Mean)
Female
teachers
(Mean)
t
p
72.20
72.04
0.09
> 0.1
In respect of the hypothesis that teachers
who are single are more committed than
those who are married, analysis revealed
that the mean commitment index of those
teachers who were married was 74.46 while
that of teachers who were single was 70.48,
which means that the married teachers were
more committed than the single teachers.
And the t-test showed that the two groups
differed from each other at a statistically
significant level, p < 0.1, t = 1.79 (see Table V).
In other words, while the fifth hypothesis
was not supported per se, it was found that
marital status and commitment were indeed
related, albeit in the opposite direction as
hypothesized, i.e. married teachers are in
fact more committed than single ones.
It was proposed that stay-back is related to
commitment. Here stay-back refers to voluntary stay-back which is made up of both
educational and non-educational stay-back.
The correlation matrix in Table VI indicated
that stay-back was positively related to
commitment (r = 0.17), meaning that those
teachers who stay-back voluntarily were,
Table V
Commitment of single teachers and married
teachers
Variable
Commitment
Single
teachers
(Mean)
Married
teachers
(Mean)
t
p
70.48
74.46
1.79
< 0.1
comparatively speaking, more committed.
However, such a correlation was weak and
not statistically significant.
Further results revealed that the correlation coefficient between commitment and
educational stay-back was 0.21 and the correlation was marginally significant, p = 0.11.
Besides, the correlation coefficient between
commitment and non-educational stay-back
was ±0.12 which means that teachers' staying
back for non-educational purposes implies
that they were less committed. However, this
correlation was found to be not statistically
significant. As shown in Table VI, educational stay-back was positively related to
commitment (r = 0.21, p = 0.11). However,
non-educational stay-back was negatively
related to commitment (r = ±0.12) but not at a
statistically significant level.
It was proposed that age is related to
commitment. Analysis showed that the two
were correlated at 0.23. This implies that the
more senior teachers, in terms of age, were a
little more committed. And the correlation
was statistically significant, p < 0.1. To
predict commitment, the following formula
shown by the line of best fit in the analysis
can be used.
Commitment = 2.80 6 Age + 67.18
Similarly, tenure was positively related to
commitment (r = 0.22) which means that the
more experienced a teacher was, the more
committed he/she would be. In this study,
tenure refers to the number of years one has
stayed in the teaching profession i.e. teaching
experience. The correlation was found to be
significant, p < 0.1 and the line of best fit was
Commitment = 0.35 6 total number of
years in teaching + 69.12
Teachers occupying higher positions in the
school are generally thought to be more
Table VI
Correlations between commitment and various independent variables ± primary analysis
1. Commitment
2. TVSI
3. TESI
4. TNSI
5. Age
6. Tenure
TVSI
TESI
TNSI
Age
Tenure
Position
0.1749
p = 0.178
0.2058
p = 0.112
0.9844
p = 0.000
0.1193
p = 0.360
0.3444
p = 0.007
±0.1740
p = 0.180
0.2330
p = 0.071
±0.0869
p = 0.505
0.0925
p = 0.478
±0.0069
p = 0.958
0.2173
p = 0.092
±0.0665
p = 0.610
±0.0713
p = 0.585
±0.0083
p = 0.950
0.8676
p = 0.000
0.0744
p = 0.569
0.0651
p = 0.618
0.0893
p = 0.494
±0.1122
p = 0.390
0.2764
p = 0.031
0.3680
p = 0.004
7. Position
[ 233 ]
committed to their job. Results also suggested
that position and commitment were positively related (r = 0.07). However, the correlation was very weak and not significant.
Using the two independent variables,
namely stay-back and position to predict
commitment, it was found that both were not
strongly related to commitment although the
former was slightly more strongly related
than was the latter (r = 0.17 vs r = 0.07).
Besides, stay-back also correlated with commitment at a comparatively more significant
level than was position (p=0.18 vs p=0.57)
though both correlations with commitment
were not significant.
Among the factors associated with commitment as examined in this study, namely
educational stay-back, non-educational stayback, total stay-back, position, tenure, gender, age and educational level, it was found
that age was the factor most strongly related
to commitment when compared with all the
other independent variables in the primary
analysis.
Since marital status, age and tenure were
found to be significantly related with commitment, these three predictor variables
were entered into the multiple regression
analysis. The best predictor chosen by the
statistic program was age because it was the
factor most highly related with commitment
among the three. But the other two predictors, marital status and tenure, were not able
to enter the computer program because there
was collinearity between them and age,
which means that marital status and tenure
measured the same underlying factor as age
and they were unlikely to improve upon the
prediction of commitment achieved by age.
In summary, the corresponding results are
shown in Table VII.
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
Discussion
In the present study, the primary research
problem involved examining the predictive
validity of the various independent variables
like educational stay-back, non-educational
stay-back, total stay-back, age, gender, tenure, position, education level, etc. on teachers' professional commitment and the
reliability of the stay-back data generated in
different ways. This was done by means of
Table VII
Analysis results for various hypotheses
Hypothesis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Result
S
X
N
N
N
N
P
N
S
S
S
N
N
N
N
Notes: S denotes the associated hypothesis is supported; X denotes the associated
hypothesis has no conclusion; N denotes the associated hypothesis is not supported; P
denotes the associated hypothesis is partially supported
[ 234 ]
correlation techniques. In addition, normality of the sample distributions for the six
stay-back indices was examined in order to
justify the use of t-test and correlation
techniques.
In the primary analysis, the first group of
variables were the stay-back indices. No
significant relationships were found between
them and commitment, except for educational stay-back which was found to be
slightly correlated with commitment (r =
0.21) at a marginally significant level (p =
0.11).
Apart from stay-back, there were other
independent variables like educational level,
marital status, age, tenure, position and
gender which were studied in this research
to find out their predictability of teachers'
professional commitment. Results revealed
that among these variables, both age and
tenure were found to be positively related
with commitment (r = 0.23 and r =0.22
respectively) at a statistically significant
level, p < 0.1. This means that the older and
the more experienced teachers were a little
more committed than those who were
younger and less experienced.
Secondary analysis
On a very lenient basis with regard to
statistical significance, it may be concluded
that educational stay-back correlates positively with commitment (r = 0.21, p = 0.11). In
other words, in the primary analysis, support
for the examined hypotheses was not at all
strong. So, a secondary analysis was undertaken.
In the secondary analysis, some hypotheses were re-analysed and re-examined using
the same commitment index, stay-back data
and personal information. However, the stayback data were manipulated, with a view to
enhancing their accuracy.
In the initial analysis, TESI was found to
be positively related to commitment at a
marginally significant level (p = 0.11). But
TVSI and TNSI did not correlate with commitment at a statistically significant level.
However, although not hypothesized,
TSIFE was found to be positively and highly
correlated with commitment (r = 0.34, p =
0.007). This finding implies that TSIFE is
quite a good predictor of teachers' professional commitment. In view of the strong
predictability of TSIFE and the inabilities of
the other stay-back indices (TVSI, TESI and
TNSI) in reflecting commitment, an attempt
was made to adjust the TVSI, TESI and TNSI
to improve their consistency with TSIFE.
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
As a matter of fact, the TSIFE was obtained
by multiplying the number of times a respondent stayed back at school in a month
with the average duration of each stay-back.
But the TESI and TNSI were generated by
aggregating the total amount of time a
respondent spent on the various stay-back
activities in a month. The TVSI is actually a
summation of the TESI and the TNSI. In the
secondary analysis, the component stay-back
indices (TVSI, TESI and TNSI) were manipulated.
In order to reduce the random scatter of the
sample distribution of stay-back time and to
correct probable errors of the TVSI, TESI and
TNSI in predicting teachers' professional
commitment, the three stay-back indices
were adjusted under the following assumptions:
1 The respondents had put down a more
correct estimate of the total aggregate
stay-back time (TSIFE) in the first part of
the questionnaire.
2 As for the second part of the questionnaire, the respondents had put down a
relatively inaccurate estimate of their
compulsory, educational and non-educational stay-back time, which made up TSI,
but
3 They had put down the correct proportion
of time they spent on each educational or
non-educational stay-back activity.
The first assumption was made because it
seems easier for a respondent to estimate the
total amount of time he or she spent after
school on each day he or she stayed back than
to recall how much time had been spent on
each individual stay-back activity since there
were so many of the latter.
In filling in the questionnaire, some respondents might have underestimated while
some overestimated the amount of time they
spent on each stay-back activity. In some rare
cases, a respondent might have committed
the mistakes of both underestimation and
overestimation of the time spent on different
stay-back activities. Thus, the second assumption was made.
Although a respondent might not have put
down the correct amount of time he or she
spent on each stay-back activity, it is quite
likely that once the time spent on one stayback activity had been overestimated, the
respondent would overestimate the time
spent on all the other activities at a constant
proportion since he or she would have made
comparisons with activities that had been
reported. The same applied in cases of
underestimation. The third assumption was
made under this reasoning. If, however, the
respondent overestimated the time spent on
half of the activities while underestimating
the other half, which would be rare anyway,
the following adjustment would offer little
help.
The TVSI, TESI and TNSI were adjusted
through the following formulae respectively:
.
Adjusted total voluntary stay-back index
(ATVSI) = TVSI 6 TSIFE/TSI
.
Adjusted total educational stay-back index
(ATESI) = TESI 6 TSIFE/TSI
.
Adjusted total non-educational stay-back
index (ATNSI) = TNSI 6 TSIFE/TSI
These three formulae served the purpose of
magnifying the overestimated stay-back
hours and reducing the underestimated ones
towards the more accurate aggregate of
TSIFE (instead of TSI), while retaining the
relative proportion of each type of stay-back
activity of each subject.
In the secondary analysis, the t-test was
employed to test some of the hypotheses.
Therefore, the normality of the adjusted stayback indices were again examined prior to
any major analysis.
All the three adjusted stay-back indices
were still found to deviate from normal
distribution. Comparatively speaking, ATNSI was furthest from normal (kurtosis = 2.02,
skew = 1.47) while ATVSI and ATESI more
closely approximated normality. The kurtosis and skew of ATVSI and ATESI were ±0.98,
0.39 and ±0.84, 0.5 respectively. In all three
indices, the sample distribution tended to be
skewed to the left. This means that the
majority of the sample had stay-back hours
less than the mean. However, the distributions of ATVSI, ATESI and ATNSI may be
considered normal as skew did not approach
2 and kurtosis was less than 5. So, the three
adjusted stay-back indices were fit for statistical analysis (Table VIII).
Hypotheses 1, 2, 6, 7, 8, 9 and 14 which
examined the relationship between commitment and the various kinds of stay-back were
studied again after the adjustment of the
stay-back indices. It is clear that, as far as the
Table VIII
Sample distribution of various adjusted
independent variables
Mean
Minimum
Maximum
Standard deviation
Kurtosis
Skewness
N
ATESI
ATNSI
ATVSI
22.66
0.00
53.41
15.33
±0.84
0.50
1220
1.97
0.00
8.65
2.16
2.02
1.47
1220
24.63
0.00
54.65
15.42
±0.98
0.39
1220
[ 235 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
nature of stay-back in school is concerned,
there was a general education bias. This
general education bias of stay-back (i.e. bias
against non-educational activities) provided
a new criterion for regrouping the subjects so
that a similar hypothesis can be examined.
Accordingly, Hypothesis 2 was modified to
become Hypothesis 2A as follows:
Teachers who are more educationally
biased than average in their stay-back
hours are more committed than those who
are less educationally biased than average.
Results for the hypothesis testing are tabulated in Table IX.
As mentioned in the analyses earlier,
ATESI, marital status, tenure and age were
significantly related with commitment; these
four factors were therefore entered into the
multiple regression analysis.
Since ATESI was the best predictor of
commitment (r = 0.289), it was the first
predictor entered by the statistic program into
the multiple regression. This procedure
yielded a multiple correlation coefficient (R)
which was the same as the correlation
coefficient (r) in the second column of Table X.
The second predictor chosen was marital
status which was not selected by the statistic
program on the basis of its correlation (r)
with the criterion variable, commitment. It
was chosen on the basis of how well it
improved on the prediction achieved by the
first variable, ATESI. As shown in Table X,
ATESI and marital status together yielded a
multiple correlation coefficient (R) of 0.426.
The third predictor of commitment entered
into the multiple regression analysis was age.
Table IX
Analysis results for various hypotheses with
adjustment
Hypothesis
1
2
6
7
8
9
14
Result
S
N
S
P
N
S
S
Notes: S denotes the associated hypothesis is
supported; X denotes the associated hypothesis has
no conclusion; N denotes the associated hypothesis is
not supported; P denotes the associated hypothesis is
partially supported
Table X
Stepwise multiple regression of influence
variables on teachers' professional
commitment
[ 236 ]
Influence
variables
Corr.
Multiple
Coef. (r) Corr. (R)
ATESI
Martial status
Age
0.289
±0.220
0.233
0.288
0.426
0.478
R2
R2Inc.
0.083
0.182 0.099
0.227 0.045
It slightly improved the prediction made by
the first two predictors to R = 0.478.
The multiple correlation coefficient (R)
mentioned above is a measure of the magnitude of relationship between a criterion
variable and a predictor variable or some
combination of predictor variables. In Table
IX, the value of R gradually increased from
0.288 to 0.478 as each predictor variable was
added. The value of 0.478 represented the best
prediction of commitment that could be made
from the influence variables listed in the
table.
The coefficients of determination (R2) were
the square of the corresponding R coefficients in the third column. They expressed
the amount of variance in the criterion
variable that is predictable from a predictor
variable or combination of predictor variables.
The R2 increments is a statistic that
expresses the additional variance in the
criterion variable that can be explained by
adding a new predictor variable to the
multiple regression analysis. In this study,
the addition of marital status to the analysis
explained 9.9 percent more of the variance in
the criterion variable (18.2 per cent-8.3 per
cent = 9.9 per cent) than could be explained
by ATESI alone. Adding age to the analysis
resulted in an R2 increment of just 4.5 per
cent, meaning that it predicted 4.5 per cent
more of the variance in level of commitment
than could be predicted by just ATESI and
marital status combined.
Although tenure was also found to be
significantly related with commitment, it
could not enter the multiple regression
analysis because there was collinearity i.e.
overlap, between age and tenure. The two
variables correlated highly with each other (r
= 0.87) because they measured the same
underlying factor. Tenure could not enter the
multiple regression analysis because it was
unlikely to improve on the prediction.
Therefore, Hypothesis 16 was quite wellsupported in the secondary analysis after the
educational stay-back index had been adjusted, resulting in one more predictor variable, ATESI, to be put into the multiple
regression analysis. The three influence
variables, namely ATESI, marital status and
age when combined yielded a very high
relationship with commitment which was
R = 0.478.
In general, results were found to be more
statistically significant after the stay-back
indices had been adjusted in the second
analysis to reflect more accurately the time
the respondents spent on each stay-back
activity. The findings showed that stay-back,
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
especially educational stay-back, could better
reflect teachers' professional commitment
and it could therefore be used as a means for
estimating commitment.
respectively. However, the three yielded a
multiple correlation coefficient (R) of 0.48
when combined.
Research implication
Conclusion
This paper was intended to investigate the
effects of stay-back on Hong Kong aided
secondary school teachers' professional
commitment. Apart from stay-back, the effects of othe
commitment
Antia Hung
St Teresa Secondary School, Kowloon, Hong Kong
James Liu
Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong
Keywords
Teachers, Commitment, Schools,
Management
Abstract
This study investigated the effects
of stay-back on teachers' professional commitment, in order to
develop an effective means to
measure teachers' commitment to
the teaching profession which,
according to various studies, contributes to ``quality education''.
Other possible factors, such as
age, gender, marital status, tenure, educational level and position, were also studied.
Suggestions on ways to increase
commitment, based on the statistical findings, were also made. The
results of causal-comparative,
correlation and multiple regression analyses revealed that stayback was the one factor which
was found to be the most highly
and significantly related to commitment. Apart from educational
stay-back, marital status, age and
tenure were also proved to relate
significantly to commitment. In
this research, the accuracy and
reliability of the stay-back data
were improved through adjustment for a secondary analysis.
The International Journal of
Educational Management
13/5 [1999] 226±240
# MCB University Press
[ISSN 0951-354X]
[ 226 ]
In Hong Kong, the study of teachers' professional commitment is of the utmost importance. People begin to question the
effectiveness of the school, especially when
juvenile delinquency seems to become a
serious problem and when academic performance seems to have fallen in the wake of
rapid expansion of secondary and tertiary
education. The situation is further aggravated by the declining teacher quality as a
result of teacher wastage.
Teacher wastage is an obstacle to quality
education. Since the 1980s, the teaching
profession in Hong Kong has been facing the
problem of teacher exodus as a result of the
political uncertainty in Hong Kong after 1997.
Government statistics show that wastage of
school teachers has been edging up over the
past few years. In secondary schools, the
wastage rate among teachers has increased
from 6 per cent in 1988 to more than 10 per
cent annually (South China Morning Post,
1990). Research carried out by the Hong Kong
Institute of Personnel Management (HKIPM)
in 1994 revealed that the education sector
provided the second highest number of
professional people leaving the territory in
the preceding two years and the number of
teachers emigrating showed a big increase.
School managers have difficulty in maintaining some committed teachers who have
been in the profession for quite some time
and this in turn makes it difficult for schools
to become effective.
The problem that the education field faces
is further exacerbated by the fact that fewer
people are willing to join the teaching
profession. From 1987-90, applications for
full-time teacher training have dropped by
about 40 per cent. This was attributed to the
lack of promotion prospects in schools,
difficult students and demanding parents, the
burden of too many non-professional duties
The current issue and full text archive of this journal is available at
http://www.emerald-library.com
and the expansion of tertiary education
which the majority of the students in Hong
Kong will put as the first priority of their
educational pursuit. After all, teaching is
regarded as a stressful job which is becoming
more unpopular and, therefore, it is less
preferred by the young in Hong Kong.
On the other hand, there is the student
problem. With the advances of society, juvenile problems have been escalating. There
are many contributing factors, for example,
material temptation, broken families, pressure from school work, the inclusion of all
school-age children in compulsory education,
etc. The new generation of students are more
rebellious and quite a number of them are
trouble-makers. They do not only do poorly
in their academic work but they also have
behavioural problems. Facing such students,
the teaching profession demands more commitment from teachers. But unfortunately,
the supply of committed teachers falls far
short of demand. Wishing to teach the cream
of the crop is the general mentality of the
teachers. Very few are willing to spare extra
time and effort to help the under- or even
average-achievers.
With reference to the situation in Hong
Kong, to arouse the commitment of teachers
is a pressing problem befalling all school
managers nowadays. It has been reported
that teachers' commitment was positively
related to leader's charisma (Bass, 1985;
Cheng, 1990). The Hong Kong Education
Department is trying to introduce reforms in
school management. One typical example is
the school management initiative (SMI) under which schools are given more flexibility
and autonomy in resource management
(Education and Manpower Branch and Education Department, 1991). Another example is
the recent introduction of quality education
fund to finance proposals which aim at
improving the quality of education (Education and Manpower Branch and Education
Department, 1998). The essence of these
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
schemes is participation and accountability.
It is believed that if teachers are involved in
the setting of school goals and the decisionmaking process, they will tend to be committed members of staff. Arousing the commitment of teachers is thus one of the
objectives of these schemes.
This paper presents an effective and economical means which can truly predict
commitment and yet is free from the limitations of the conventional methods of measuring commitment that meet with
employees' suspicion of the motives of the
management, as this new approach can be
applied without employees' knowing it. It
focuses on investigating the correlation of
teachers' stay-back to their professional
commitment.
Commitment and contributing
factors
By professional commitment, we mean teachers' commitment to their job as a teacher.
It corresponds to commitment to work put
forward by Fink (1992) which is related to
organization, co-worker and work.
Since teaching is a profession comprising a
considerable number of members, conceptually we can see the teaching profession as a
large organization. Based on an analogy to
organization commitment defined by Mowday and McDade (1979), we characterize
professional commitment by the following
factors:
.
a strong belief in and acceptance of the
profession's goals and values;
.
a willingness to exert considerable effort
on behalf of the profession; and
.
a strong desire to maintain membership
in the profession.
Numerous studies have suggested that employees' job attitudes and commitment can
affect their performance (Cammann et al.,
1983; Oldham and Hackman, 1981). For personal factors, they include locus of control
(Cheng, 1990), age, tenure, gender and education level (Fink, 1992). Non-demographic
individual difference variables include central life interest (Dubin et al., 1975), desire for
greater job responsibility, expectations about
goal achievement and personal needs (Steers,
1977).
For organizational characteristics, they
consist of job, leadership, work group and
organizational characteristics. Among these
four characteristics, leadership and organizational characteristics seem to be of utmost
concern to school managers who want to
arouse the commitment and enhance the
performance of teachers. Leadership refers to
the styles of the leaders which can be supervisory (Fukami and Larson, 1984), charismatic (Conger et al., 1988), one of initiating
structure and consideration (Morris and
Sherman, 1981) and one of giving rewards
(Bateman and Strasser, 1984). For organizational characteristics, they include things
like formalization, decentralization, functional dependence on the work of others and
participation in decision making (Morris and
Steers, 1980; Steers and Rhodes, 1978).
Other factors such as lateness, absenteeism, tardiness and turnover are the manifestations of commitment as reflected in most
literature. But there is a kind of behaviour
which can be a predictor of employees' level
of commitment and which has not yet been
studied thoroughly. It is ``employees' stayback time after normal office hours''. As
discussed before, leadership style is an
important factor affecting commitment. By
providing teachers with inspiration, encouragement and more meaning to their work, a
charismatic principal can enhance a teacher's faith in and respect for him and these
lead to an increase in the teacher's commitment to the principal and so to the school and
the work. Additional studies by Cole (1979)
and Sekaran (1989) have further reinforced
our belief that employees' ``stay-back'' is also
a manifestation of commitment and can be
used as a predictor of commitment level,
apart from the traditional commitment-related behaviour mentioned by other researchers.
In short, it is suggested that a school
manager can generate commitment from his
employees by building trust, letting people
develop their own ways of working, sharing
accountability and ownership of a job, and
negotiating help and supervision in terms
that engender employee development. But
above all, managers need to be role models
for their subordinates by being committed
and they should also empower others in their
jobs and roles.
Traditional measures of
commitment
There are two approaches to measure commitment: formal and informal. The formal
approach refers to the administering of
questionnaires to employees to measure their
commitment. Two commonly-used questionnaires are the ``organizational commitment
questionnaire'' devised by Mowday and
McDade (1979) and the ``commitment diagnosis instrument'' by Fink (1992). However, the
[ 227 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
formal method is often met with resistance
from the employees, especially in an organization where there is a lack of trust from the
management. Employees may suspect the
motives of the management. Instead of being
viewed as a means of understanding the
organization and the needs of its employees,
the questionnaire can easily be considered as
a form of employee evaluation that leads to
judgments about the quality of the employees
themselves, which can form the basis for
promotion, demotion or dismissal. As a
result, employees exercise extra caution, not
because they understand that they must be
honest but because they are more concerned
about how they should respond to the questions. Even attempts to guarantee anonymity
can be futile. The results generated from the
formal method may not be a true picture of
employee commitment.
The informal approach employs the observational techniques like commitment indicators which are actually commitmentrelated behaviour, for instance, absenteeism,
lateness, turnover, etc. Although absenteeism is the most widely studied predictor, it is
not an accurate and reliable one as most
reviewers of the literature on absenteeism
argue that current research abounds in
problems of inadequate theory, inaccurate
measurement, invalid data and faulty analytic procedures (Goodman et al., 1984).
Stay-back
In Hong Kong, all the aided secondary
schools are full-time. Recesses and lunch
time are very short, totalling less than two
hours each day. Teachers have a very heavy
workload. On average, a teacher has to teach
six periods out of eight every day. Apart from
teaching, teachers also have to perform some
non-teaching duties, for example, they may
have to be on duty during recess or lunch
time. So, they have little contact with students and colleagues during school hours.
Counselling work, extra-curricular activities, the marking of assignments, socializing, etc. will have to be carried out after
school hours. Without staying back, it is
quite unlikely that teachers can accomplish
their tasks and perform well.
As a commitment predictor, stay-back can
be defined as the time that employees stay
behind in the office after the formal and
scheduled office hours, regardless of what
they do (Cole, 1979; Sekaran, 1989). In this
study, the employees referred to the aided
secondary school teachers in Hong Kong
randomly selected for this research purpose
[ 228 ]
and stay-back was broadly divided into two
types, namely voluntary and compulsory
stay-back. Voluntary stay-back happens out
of the teachers' own free will. On the other
hand, compulsory stay-back, as the name
suggests itself, happens because teachers are
required to stay as part of the fulfilment of
their official duties.
In addition, voluntary stay-back can
further be divided into two sub-groups:
educational stay-back and non-educational
stay-back. Figures 1 and 2 give the classifications of the various stay-back activities.
Data collection
The present study was conducted among
aided secondary school teachers in Hong
Kong. In the first stage, questionnaires were
distributed directly to about 5 per cent of
aided secondary schools in Hong Kong randomly chosen for this research. A return rate
of 45 per cent was recorded. However, among
these questionnaires, one third did not include complete data, thus making analysis
difficult and were then discarded. In order to
improve the reliability of the study, another
batch of questionnaires was given out in the
second stage at the Professional Teachers'
Union where teachers used to go shopping.
Teachers were randomly intercepted and
teachers in government aided and subsidized
secondary schools were asked to fill in the
questionnaire. The response rate was quite
reasonable and both stages managed to
collect a total useful sample of 1,220 representing about 5 per cent of teacher population
in aided secondary schools.
In the questionnaire, teachers were asked
questions about their general feelings towards the teaching profession, matters pertaining to their stay-back, and personal
information such as age, gender, marital
status, position and teaching experience.
Figure 1
Classifications of activities in compulsory stayback
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
Figure 2
Classifications of activities in voluntary stay-back
The International Journal of
Educational Management
13/5 [1999] 226±240
Teachers' frequency of stay-back and the
total stay-back hours per month were calculated from the information that they gave in
the questionnaire. The dependent variable
(teachers' professional commitment) was regressed on the independent variables (stayback indices, age, gender, marital status,
tenure and position). In addition, the characteristics of stay-back data were analysed
and investigated.
Sample design and analysis
The sample consists of 1,220 teachers in Hong
Kong who were randomly chosen from some
aided secondary schools and from those who
visited the Professional Teachers' Union. In
terms of gender, the subjects were quite
equally distributed, 48 per cent male and 52
per cent female. The average age of those in
the sample was 35, with the 21-30 age group
accounting for 42 per cent, the 31-40 group 46
per cent and the 41-50 group 12 per cent. Fiftysix per cent of the subjects were single and 44
per cent were married.
As far as education is concerned, 65 per
cent of the subjects held bachelor's degrees or
above while 35 per cent were non-graduates
who might only have a teacher's certificate.
In Hong Kong, the non-graduates start
teaching as Certificated Masters/Mistresses
(CM), then they may be promoted to assistant
masters/mistresses (AM). Some outstanding
performers may then be further promoted to
senior assistant masters/mistresses (SAM)
and finally to principal assistant masters/
mistresses (PAM), which, at present, is a
position held by one only in every aided
secondary school with an average of 50
teachers (see Figure 3).
In the stream for graduates, they start as
graduate masters/mistresses (GM). Some
may be promoted to senior graduate masters/
mistresses (SGM). Further promotion will
make them end up as principal graduate
masters/mistresses(PGM) who are actually
deputy principals. At present, there are two
PGMs in every aided secondary school.
Among the sample in the present study, 28
per cent are CMs, 5 per cent AMs, 42 per cent
GMs, 22 per cent SGMs, and 3 per cent PGMs,
i.e. 31 per cent of the respondents had been
promoted to senior positions. When teaching
experience is concerned, 35 per cent had
taught for five years or less while those who
had taught between six to ten years constituted 31 per cent, 11 to 15 years 25 per cent, 1620 years 7 per cent and only 2 per cent had
taught for more than 20 years (see Figure 4).
The questionnaire documented the frequency and the duration of and the reasons
for the teachers' stay-back. All respondents
were asked to put down on average how often
they stayed back each month and the average
Figure 3
Distribution of professionals
[ 229 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
Figure 4
Distribution of teaching experience
The International Journal of
Educational Management
13/5 [1999] 226±240
amount of time for each stay-back. Some
typical stay-back activities, which can
broadly be divided into two types: voluntary
and compulsory, were also listed. These data
can be used to cross check the total amount of
time a respondent spent on the stay-back
activities in an average month, regardless of
the nature of stay-back, be it voluntary or
compulsory. To prevent any patterns that
might lead a subject to automatically circle
numbers down one side or the other of the
seven-point scale used in the questionnaire,
some of the items in the question set are
stated positively, where a higher score reflects greater commitment; while some of the
items are stated negatively, where a higher
score reflects less commitment. Also the
positive and negative statements were mixed
randomly, not systematically, again to avoid
any pattern. Items were summed and converted to a 10-100 scale to generate the
commitment index for each subject.
Stay-back indices
The study employed six stay-back indices as
follows:
1 Total stay-back index by frequency estimate (TSIFE).
2 Total compulsory stay-back index (TCSI).
3 Total educational stay-back index (TESI).
4 Total non-educational stay-back index
(TNSI).
5 Total voluntary stay-back index (TVSI).
6 Total stay-back index (TSI).
The TSIFE was generated by multiplying the
total number of stay-back days each month
with the average amount of time for each
stay-back, regardless of the reasons for stayback, where the number of stay-back days
each month was calculated from the frequency (once how many days) of stay-back,
assuming 20 school days per month.
The TCSI only counted those activities
which had been classified as compulsory
stay-back activities in the last section. Since
the respondents had specified the number of
times they carried out a certain compulsory
[ 230 ]
stay-back activity in an average month and
the duration each time, the total amount of
time spent on a certain activity in a month
can be found. The TCSI is actually a summation of all the time spent on compulsory
stay-back activities by a respondent in a
month.
The TESI counted those activities classified as voluntary educational stay-back activities. Again the total hours spent on these
activities by a respondent in an average
month was a summation of the time spent on
each voluntary educational stay-back each
month which was calculated by multiplying
the number of times with the duration each
time. As discussed in the last section, some of
the activities listed in the questionnaire can
actually be grouped under either educational
or non-educational. In the data analysis, half
of the total amount of time spent on this
group of activities was counted in the TESI
while the other half of it was counted in the
TNSI.
The TNSI counted those activities defined
as voluntary non-educational stay-back activities. It is the summation of the total
amount of time spent on each of these
activities by a respondent in an average
month which was calculated by multiplying
the number of times with the duration each
time. Like the TESI, the TNSI also included
half of the total amount of time spent on
activities which can be counted as either
educational or non-educational. The TVSI is
the sum of the TESI and the TNSI. The TSI is
the sum of the TCSI, TESI and TNSI.
In theory, the TSIFE should equal the TSI if
the respondents were absolutely consistent
in answering the questionnaire. The built-in
cross-checking device during data collection
is a means to ensure reliability of data.
Hypotheses testing
Basically, the present research examined
some hypotheses using statistical techniques
such as differences in means and standard
deviations, t-test, F-test correlations, and
multiple regression model.
Before testing the hypotheses, the consistency and reliability of the stay-back data
generated from frequency estimate and that
obtained by an aggregation of the total
compulsory stay-back hours, total educational stay-back hours and total non-educational stay-back hours were investigated. The
two kinds of information were related to each
other and their correlation coefficient was
calculated.
Apart from the above method, the following was also employed in the data analysis.
Stay-back data were analysed by each stay-
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
back index to examine whether or not its
sample distributional characteristics approximate normality. The same has been
done for variables like age, position and
tenure. Specifically, Kendall and Stuart's
criteria were used to examine normality of
sample distributions. Those independent
variables with normal distributional patterns could then be utilized in the multiple
regression analysis.
Hypothesis 1
Teachers who stayed back voluntarily were
distinguished from those who did not by
forming two groups. Using the total voluntary
stay-back hours (both educational and noneducational), the first group was formed by
subjects with total voluntary stay-back hours
less than the median, while the second group
was subjects with total voluntary stay-back
hours more than the median. Then, the means
and the standard deviations of the commitment indices of the two groups were calculated.
A t-test was conducted to see the significance of
the difference between the two groups and the
correlation coefficients between commitment
and stay-back were compared.
Hypothesis 2
Teachers who stayed back voluntarily were
divided into two groups according to the
nature of their stay-back. The first group
consisted of those who spent 51 per cent or
more of the total voluntary stay-back time on
educational stay-back activities while the
second group comprised those who spent 51
per cent or more of their total voluntary stayback time on non-educational stay-back activities. Then, the means and the standard
deviations of the commitment indices of the
two groups were calculated. A t-test was
conducted to see the significance of the
difference between the two groups and the
correlation coefficients between commitment
and educational stay-back and that between
non-educational stay-back were compared.
Hypotheses 3, 4 and 5
Teachers were divided into two groups
according to their educational level, gender
and marital status, as required in Hypotheses
3, 4 and 5 respectively. In Hypothesis 3, the two
groups of teachers were graduates and nongraduates. In Hypothesis 5, teachers who were
single and those who were married made up the
two groups. Under each hypothesis, the means
and the standard deviations of the commitment indices of the two groups were calculated.
A t-test was conducted to see the significance of
the difference between the two groups. The
correlation coefficients between commitment
and educational level, as in Hypothesis 3, that
between commitment and gender, as in
Hypothesis 4 and that between commitment
and marital status, as in Hypothesis 5 were
compared.
Hypotheses 6-14
Teachers' professional commitment was predicted by working out the correlation coefficient between each subject's commitment
and each of the independent variables such
as TVSI, TESI, TNSI, age, gender, marital
status, tenure, position and educational level.
This was presented in the form of a correlation matrix. A line of best fit, y = bx + a, was
generated, to facilitate the necessary prediction, when and where the relationship was
significant and amply correlated.
Hypothesis 15
The few most influential predictor variables
were used for a 3-4 step multiple regression
analysis. The resulting multiple correlation
coefficients (R) were examined for their
magnitudes.
Statistical analysis
Findings of the analysis of the hypotheses
include two main streams: the examination
of the stay-back data and the prediction of
professional commitment using various independent variables. The primary concern of
the first stream was to test stability of the
stay-back data along with other statistical
attributes (i.e. normality) associated with the
six stay-back indices. The second stream
focused on validating the use of voluntary
stay-back indices and other independent
variables like age, gender, tenure and educational level to predict professional commitment.
Consistency and reliability of stay-back
indices
It was found that TSIFE and TSI were
correlated with each other and gave a
correlation of 0.48, significant at p = 0.01
level. Therefore, it can be concluded that the
two indices are highly and positively correlated and the respondents were consistent in
giving their stay-back data. Since the other
stay-back indices such as TCSI, TESI and
TNSI are complements of the TSI, like the TSI
and TSIFE, they are reliable indices for
statistical analysis in the present research.
Normality of the stay-back indices
Hypotheses 1 to 5 were tested by means of
causal-comparative method. The t-test is
probably the most commonly used statistical
tool in causal-comparative studies and it has
made three assumptions about the scores
[ 231 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
obtained in research. One is that the scores
in the population under study are normally
distributed. So, it was necessary to test the
normality of the stay-back data before any
statistical analysis was carried out.
The characteristics of each sample distribution were examined using measures of
skew and kurtosis, as well as pictorial
shapes. Statistically, distributions may be
considered non-normal, indicating the presence of skew and leptokurtosis, when the
skew approaches 2 and kurtosis exceeds 5. In
normal distribution, both skew and kurtosis
are zero. If the value of the skew is positive,
distribution is skewed to the left but if its
value is negative, the distribution is skewed
to the right. For kurtosis, its value determines the peak of the distribution. If kurtosis
is greater than zero, the distribution is more
peaked than normal but if kurtosis is less
than zero, the shape of the curve is flatter
than the normal curve.
In this study, all six stay-back indices were
found to deviate from normal distribution.
According to Kendall and Stuart's criterion
for normality, the distributions of TCSI and
TNSI seemed to be furthest from normal. As
shown in Table I, kurtosis and skew of the
former are 3.88 and 1.77 respectively while
that of the latter are 3.2 and 1.83 respectively.
The other four stay-back indices, namely
TESI, (kurtosis = ±0.04 and skew = 0.83), TSI,
(kurtosis = ±0.59 and skew = 0.4), TVSI,
(kurtosis = ±0.27 and skew = 0.74) and TSIFE,
(kurtosis = ±1.23 and skew = ±0.2) more
closely approximated normality. In all the
four indices, the distribution curve is flatter
than the normal one as the values of kurtosis
are less than zero and the sample distribution tends to be skewed to the left in the cases
of TESI, TSI and TVSI which means that the
majority of the sample had stay-back hours
less than the mean.
Although the six stay-back indices do not
have perfectly normal distribution, they are
within an acceptable range of normal distribution with different magnitudes of deviations. They, therefore, are fit for a t-test
analysis.
[ 232 ]
In reporting the results of the analysis, the
significance level was taken at p 4 0.1. It was
proposed that teachers who stay back are
more committed than those who do not. The
analysis revealed that the mean commitment
index of teachers who stayed back was a bit
higher than that of teachers who did not (72.2
vs 72). Nevertheless, the difference was not
statistically significant (see Table II).
The analysis for the hypothesis that teachers who stay back more for educational
purposes are more committed than those who
stay back more for non-educational purposes
was found to be infeasible because there was
a great difference in the number of subjects
between the two groups of teachers. Out of
the 1,220 subjects, only 61 (i.e. 5 per cent)
stayed back more for non-educational purposes, while 1,139 (i.e. 93.4 per cent) have
longer stay-back hours for educational purposes than non-educational ones. Twenty
respondents (i.e. 1.6 per cent) reported no
stay-back of any kind. So, a comparison could
not be made.
For the non-graduates who were posited to
be more committed teachers than graduates,
results showed that the mean commitment
index of graduate teachers was 73.34 and that
for non-graduate teachers was 71.64. This
finding contradicted the hypothesis but such
a finding is not statistically significant (see
Table III).
It was proposed that women are more
committed teachers than men. When the
commitment indices of the two groups were
analysed, the mean commitment index of
men was a bit higher than that of women
(72.20 Vs 72.04). But the t-test demonstrated
that the two groups were not different from
each other in terms of commitment at a
statistically significantly level (see Table IV).
Table II
Commitment of teachers who stay back and
teachers who do not
Variable
Table I
Sample distribution of various independent variables
Mean
Minimum
Maximum
Standard deviation
Kurtosis
Skewness
N
Results
Commitment
TCSI
TNSI
TESI
TSI
TVSI
TSIFE
10.84
0.00
48.50
9.32
3.88
1.77
1220
2.89
0.00
15.43
3.6
3.2
1.83
1220
27.22
0.00
74.00
19.23
±0.04
0.83
1220
40.95
1.90
87.00
21.29
±0.59
0.40
1220
30.11
0.00
74.00
20.18
±0.27
0.74
1220
35.06
0.91
60.00
17.36
±1.23
±0.20
1220
Teachers Teachers not
stay-back
stay-back
(Mean)
(Mean)
72.20
72.00
t
p
0.57 >0.1
Table III
Commitment of graduate teachers and nongraduate teachers
Variable
Commitment
Graduate Non-graduate
teachers
teachers
(Mean)
(Mean)
73.34
72.00
t
p
±0.3 > 0.1
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
Table IV
Commitment of male teachers and female
teachers
Variable
Commitment
Male
teachers
(Mean)
Female
teachers
(Mean)
t
p
72.20
72.04
0.09
> 0.1
In respect of the hypothesis that teachers
who are single are more committed than
those who are married, analysis revealed
that the mean commitment index of those
teachers who were married was 74.46 while
that of teachers who were single was 70.48,
which means that the married teachers were
more committed than the single teachers.
And the t-test showed that the two groups
differed from each other at a statistically
significant level, p < 0.1, t = 1.79 (see Table V).
In other words, while the fifth hypothesis
was not supported per se, it was found that
marital status and commitment were indeed
related, albeit in the opposite direction as
hypothesized, i.e. married teachers are in
fact more committed than single ones.
It was proposed that stay-back is related to
commitment. Here stay-back refers to voluntary stay-back which is made up of both
educational and non-educational stay-back.
The correlation matrix in Table VI indicated
that stay-back was positively related to
commitment (r = 0.17), meaning that those
teachers who stay-back voluntarily were,
Table V
Commitment of single teachers and married
teachers
Variable
Commitment
Single
teachers
(Mean)
Married
teachers
(Mean)
t
p
70.48
74.46
1.79
< 0.1
comparatively speaking, more committed.
However, such a correlation was weak and
not statistically significant.
Further results revealed that the correlation coefficient between commitment and
educational stay-back was 0.21 and the correlation was marginally significant, p = 0.11.
Besides, the correlation coefficient between
commitment and non-educational stay-back
was ±0.12 which means that teachers' staying
back for non-educational purposes implies
that they were less committed. However, this
correlation was found to be not statistically
significant. As shown in Table VI, educational stay-back was positively related to
commitment (r = 0.21, p = 0.11). However,
non-educational stay-back was negatively
related to commitment (r = ±0.12) but not at a
statistically significant level.
It was proposed that age is related to
commitment. Analysis showed that the two
were correlated at 0.23. This implies that the
more senior teachers, in terms of age, were a
little more committed. And the correlation
was statistically significant, p < 0.1. To
predict commitment, the following formula
shown by the line of best fit in the analysis
can be used.
Commitment = 2.80 6 Age + 67.18
Similarly, tenure was positively related to
commitment (r = 0.22) which means that the
more experienced a teacher was, the more
committed he/she would be. In this study,
tenure refers to the number of years one has
stayed in the teaching profession i.e. teaching
experience. The correlation was found to be
significant, p < 0.1 and the line of best fit was
Commitment = 0.35 6 total number of
years in teaching + 69.12
Teachers occupying higher positions in the
school are generally thought to be more
Table VI
Correlations between commitment and various independent variables ± primary analysis
1. Commitment
2. TVSI
3. TESI
4. TNSI
5. Age
6. Tenure
TVSI
TESI
TNSI
Age
Tenure
Position
0.1749
p = 0.178
0.2058
p = 0.112
0.9844
p = 0.000
0.1193
p = 0.360
0.3444
p = 0.007
±0.1740
p = 0.180
0.2330
p = 0.071
±0.0869
p = 0.505
0.0925
p = 0.478
±0.0069
p = 0.958
0.2173
p = 0.092
±0.0665
p = 0.610
±0.0713
p = 0.585
±0.0083
p = 0.950
0.8676
p = 0.000
0.0744
p = 0.569
0.0651
p = 0.618
0.0893
p = 0.494
±0.1122
p = 0.390
0.2764
p = 0.031
0.3680
p = 0.004
7. Position
[ 233 ]
committed to their job. Results also suggested
that position and commitment were positively related (r = 0.07). However, the correlation was very weak and not significant.
Using the two independent variables,
namely stay-back and position to predict
commitment, it was found that both were not
strongly related to commitment although the
former was slightly more strongly related
than was the latter (r = 0.17 vs r = 0.07).
Besides, stay-back also correlated with commitment at a comparatively more significant
level than was position (p=0.18 vs p=0.57)
though both correlations with commitment
were not significant.
Among the factors associated with commitment as examined in this study, namely
educational stay-back, non-educational stayback, total stay-back, position, tenure, gender, age and educational level, it was found
that age was the factor most strongly related
to commitment when compared with all the
other independent variables in the primary
analysis.
Since marital status, age and tenure were
found to be significantly related with commitment, these three predictor variables
were entered into the multiple regression
analysis. The best predictor chosen by the
statistic program was age because it was the
factor most highly related with commitment
among the three. But the other two predictors, marital status and tenure, were not able
to enter the computer program because there
was collinearity between them and age,
which means that marital status and tenure
measured the same underlying factor as age
and they were unlikely to improve upon the
prediction of commitment achieved by age.
In summary, the corresponding results are
shown in Table VII.
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
Discussion
In the present study, the primary research
problem involved examining the predictive
validity of the various independent variables
like educational stay-back, non-educational
stay-back, total stay-back, age, gender, tenure, position, education level, etc. on teachers' professional commitment and the
reliability of the stay-back data generated in
different ways. This was done by means of
Table VII
Analysis results for various hypotheses
Hypothesis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Result
S
X
N
N
N
N
P
N
S
S
S
N
N
N
N
Notes: S denotes the associated hypothesis is supported; X denotes the associated
hypothesis has no conclusion; N denotes the associated hypothesis is not supported; P
denotes the associated hypothesis is partially supported
[ 234 ]
correlation techniques. In addition, normality of the sample distributions for the six
stay-back indices was examined in order to
justify the use of t-test and correlation
techniques.
In the primary analysis, the first group of
variables were the stay-back indices. No
significant relationships were found between
them and commitment, except for educational stay-back which was found to be
slightly correlated with commitment (r =
0.21) at a marginally significant level (p =
0.11).
Apart from stay-back, there were other
independent variables like educational level,
marital status, age, tenure, position and
gender which were studied in this research
to find out their predictability of teachers'
professional commitment. Results revealed
that among these variables, both age and
tenure were found to be positively related
with commitment (r = 0.23 and r =0.22
respectively) at a statistically significant
level, p < 0.1. This means that the older and
the more experienced teachers were a little
more committed than those who were
younger and less experienced.
Secondary analysis
On a very lenient basis with regard to
statistical significance, it may be concluded
that educational stay-back correlates positively with commitment (r = 0.21, p = 0.11). In
other words, in the primary analysis, support
for the examined hypotheses was not at all
strong. So, a secondary analysis was undertaken.
In the secondary analysis, some hypotheses were re-analysed and re-examined using
the same commitment index, stay-back data
and personal information. However, the stayback data were manipulated, with a view to
enhancing their accuracy.
In the initial analysis, TESI was found to
be positively related to commitment at a
marginally significant level (p = 0.11). But
TVSI and TNSI did not correlate with commitment at a statistically significant level.
However, although not hypothesized,
TSIFE was found to be positively and highly
correlated with commitment (r = 0.34, p =
0.007). This finding implies that TSIFE is
quite a good predictor of teachers' professional commitment. In view of the strong
predictability of TSIFE and the inabilities of
the other stay-back indices (TVSI, TESI and
TNSI) in reflecting commitment, an attempt
was made to adjust the TVSI, TESI and TNSI
to improve their consistency with TSIFE.
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
As a matter of fact, the TSIFE was obtained
by multiplying the number of times a respondent stayed back at school in a month
with the average duration of each stay-back.
But the TESI and TNSI were generated by
aggregating the total amount of time a
respondent spent on the various stay-back
activities in a month. The TVSI is actually a
summation of the TESI and the TNSI. In the
secondary analysis, the component stay-back
indices (TVSI, TESI and TNSI) were manipulated.
In order to reduce the random scatter of the
sample distribution of stay-back time and to
correct probable errors of the TVSI, TESI and
TNSI in predicting teachers' professional
commitment, the three stay-back indices
were adjusted under the following assumptions:
1 The respondents had put down a more
correct estimate of the total aggregate
stay-back time (TSIFE) in the first part of
the questionnaire.
2 As for the second part of the questionnaire, the respondents had put down a
relatively inaccurate estimate of their
compulsory, educational and non-educational stay-back time, which made up TSI,
but
3 They had put down the correct proportion
of time they spent on each educational or
non-educational stay-back activity.
The first assumption was made because it
seems easier for a respondent to estimate the
total amount of time he or she spent after
school on each day he or she stayed back than
to recall how much time had been spent on
each individual stay-back activity since there
were so many of the latter.
In filling in the questionnaire, some respondents might have underestimated while
some overestimated the amount of time they
spent on each stay-back activity. In some rare
cases, a respondent might have committed
the mistakes of both underestimation and
overestimation of the time spent on different
stay-back activities. Thus, the second assumption was made.
Although a respondent might not have put
down the correct amount of time he or she
spent on each stay-back activity, it is quite
likely that once the time spent on one stayback activity had been overestimated, the
respondent would overestimate the time
spent on all the other activities at a constant
proportion since he or she would have made
comparisons with activities that had been
reported. The same applied in cases of
underestimation. The third assumption was
made under this reasoning. If, however, the
respondent overestimated the time spent on
half of the activities while underestimating
the other half, which would be rare anyway,
the following adjustment would offer little
help.
The TVSI, TESI and TNSI were adjusted
through the following formulae respectively:
.
Adjusted total voluntary stay-back index
(ATVSI) = TVSI 6 TSIFE/TSI
.
Adjusted total educational stay-back index
(ATESI) = TESI 6 TSIFE/TSI
.
Adjusted total non-educational stay-back
index (ATNSI) = TNSI 6 TSIFE/TSI
These three formulae served the purpose of
magnifying the overestimated stay-back
hours and reducing the underestimated ones
towards the more accurate aggregate of
TSIFE (instead of TSI), while retaining the
relative proportion of each type of stay-back
activity of each subject.
In the secondary analysis, the t-test was
employed to test some of the hypotheses.
Therefore, the normality of the adjusted stayback indices were again examined prior to
any major analysis.
All the three adjusted stay-back indices
were still found to deviate from normal
distribution. Comparatively speaking, ATNSI was furthest from normal (kurtosis = 2.02,
skew = 1.47) while ATVSI and ATESI more
closely approximated normality. The kurtosis and skew of ATVSI and ATESI were ±0.98,
0.39 and ±0.84, 0.5 respectively. In all three
indices, the sample distribution tended to be
skewed to the left. This means that the
majority of the sample had stay-back hours
less than the mean. However, the distributions of ATVSI, ATESI and ATNSI may be
considered normal as skew did not approach
2 and kurtosis was less than 5. So, the three
adjusted stay-back indices were fit for statistical analysis (Table VIII).
Hypotheses 1, 2, 6, 7, 8, 9 and 14 which
examined the relationship between commitment and the various kinds of stay-back were
studied again after the adjustment of the
stay-back indices. It is clear that, as far as the
Table VIII
Sample distribution of various adjusted
independent variables
Mean
Minimum
Maximum
Standard deviation
Kurtosis
Skewness
N
ATESI
ATNSI
ATVSI
22.66
0.00
53.41
15.33
±0.84
0.50
1220
1.97
0.00
8.65
2.16
2.02
1.47
1220
24.63
0.00
54.65
15.42
±0.98
0.39
1220
[ 235 ]
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
nature of stay-back in school is concerned,
there was a general education bias. This
general education bias of stay-back (i.e. bias
against non-educational activities) provided
a new criterion for regrouping the subjects so
that a similar hypothesis can be examined.
Accordingly, Hypothesis 2 was modified to
become Hypothesis 2A as follows:
Teachers who are more educationally
biased than average in their stay-back
hours are more committed than those who
are less educationally biased than average.
Results for the hypothesis testing are tabulated in Table IX.
As mentioned in the analyses earlier,
ATESI, marital status, tenure and age were
significantly related with commitment; these
four factors were therefore entered into the
multiple regression analysis.
Since ATESI was the best predictor of
commitment (r = 0.289), it was the first
predictor entered by the statistic program into
the multiple regression. This procedure
yielded a multiple correlation coefficient (R)
which was the same as the correlation
coefficient (r) in the second column of Table X.
The second predictor chosen was marital
status which was not selected by the statistic
program on the basis of its correlation (r)
with the criterion variable, commitment. It
was chosen on the basis of how well it
improved on the prediction achieved by the
first variable, ATESI. As shown in Table X,
ATESI and marital status together yielded a
multiple correlation coefficient (R) of 0.426.
The third predictor of commitment entered
into the multiple regression analysis was age.
Table IX
Analysis results for various hypotheses with
adjustment
Hypothesis
1
2
6
7
8
9
14
Result
S
N
S
P
N
S
S
Notes: S denotes the associated hypothesis is
supported; X denotes the associated hypothesis has
no conclusion; N denotes the associated hypothesis is
not supported; P denotes the associated hypothesis is
partially supported
Table X
Stepwise multiple regression of influence
variables on teachers' professional
commitment
[ 236 ]
Influence
variables
Corr.
Multiple
Coef. (r) Corr. (R)
ATESI
Martial status
Age
0.289
±0.220
0.233
0.288
0.426
0.478
R2
R2Inc.
0.083
0.182 0.099
0.227 0.045
It slightly improved the prediction made by
the first two predictors to R = 0.478.
The multiple correlation coefficient (R)
mentioned above is a measure of the magnitude of relationship between a criterion
variable and a predictor variable or some
combination of predictor variables. In Table
IX, the value of R gradually increased from
0.288 to 0.478 as each predictor variable was
added. The value of 0.478 represented the best
prediction of commitment that could be made
from the influence variables listed in the
table.
The coefficients of determination (R2) were
the square of the corresponding R coefficients in the third column. They expressed
the amount of variance in the criterion
variable that is predictable from a predictor
variable or combination of predictor variables.
The R2 increments is a statistic that
expresses the additional variance in the
criterion variable that can be explained by
adding a new predictor variable to the
multiple regression analysis. In this study,
the addition of marital status to the analysis
explained 9.9 percent more of the variance in
the criterion variable (18.2 per cent-8.3 per
cent = 9.9 per cent) than could be explained
by ATESI alone. Adding age to the analysis
resulted in an R2 increment of just 4.5 per
cent, meaning that it predicted 4.5 per cent
more of the variance in level of commitment
than could be predicted by just ATESI and
marital status combined.
Although tenure was also found to be
significantly related with commitment, it
could not enter the multiple regression
analysis because there was collinearity i.e.
overlap, between age and tenure. The two
variables correlated highly with each other (r
= 0.87) because they measured the same
underlying factor. Tenure could not enter the
multiple regression analysis because it was
unlikely to improve on the prediction.
Therefore, Hypothesis 16 was quite wellsupported in the secondary analysis after the
educational stay-back index had been adjusted, resulting in one more predictor variable, ATESI, to be put into the multiple
regression analysis. The three influence
variables, namely ATESI, marital status and
age when combined yielded a very high
relationship with commitment which was
R = 0.478.
In general, results were found to be more
statistically significant after the stay-back
indices had been adjusted in the second
analysis to reflect more accurately the time
the respondents spent on each stay-back
activity. The findings showed that stay-back,
Antia Hung and James Liu
Effects of stay-back on
teachers' professional
commitment
The International Journal of
Educational Management
13/5 [1999] 226±240
especially educational stay-back, could better
reflect teachers' professional commitment
and it could therefore be used as a means for
estimating commitment.
respectively. However, the three yielded a
multiple correlation coefficient (R) of 0.48
when combined.
Research implication
Conclusion
This paper was intended to investigate the
effects of stay-back on Hong Kong aided
secondary school teachers' professional
commitment. Apart from stay-back, the effects of othe